Single-stage category-level object pose estimation
AssigneeNVIDIA Corporation
InventorsStanley Thomas Birchfield, Jonathan Tremblay, Yunzhi Lin, Stephen Walter Tyree
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques to determine a pose and relative dimensions of an object from an image. In at least one embodiment, a pose and relative dimensions of an object are determined from an image based at least in part on, for example, features of the image.
Claims29
A method to determine a pose and relative dimensions of an object depicted in an image, the image having a set of features, the method comprising:
- processing the set of features to determine a first set of data corresponding to the object;
- calculating a second set of data based at least in part on the set of features and the first set of data;
- calculating a set of values corresponding to the relative dimensions of the object based at least in part on the second set of data;
- and
- processing the first set of data, the second set of data, and the set of values to determine the pose and the relative dimensions of the object depicted in the image.
A system, comprising:
- one or more computers having one or more processors to train one or more neural networks to:
- obtain an image depicting an object of a category of objects;
- determine one or more features based at least in part on the image;
- and
- predict, based on the one or more features, a set of values indicating at least a location and relative dimensions of the object in the image.
A processor, comprising:
- one or more circuits to use one or more neural networks to:
- obtain a set of features corresponding to an image;
- calculate a set of data based on the set of features, wherein the set of data of one or more bounding boxes of an object depicted in the image;
- and
- determine at least a width value, a height value, and a length value for the object based at least in part on the set of data.
A non-transitory computer-readable storage medium having stored thereon executable instructions that, as a result of being executed by one or more processors of a computer system, cause the computer system to at least:
- obtain an image depicting an object of a first category;
- generate one or more features based at least in part on the image;
- use one or more neural networks to process the one or more features to determine at least one or more coordinate values and one or more relative dimension values of the object;
- and
- process the one or more coordinate values and the one or more relative dimension values using one or more algorithms to calculate a six degrees-of-freedom (6DOF) pose of the object.
Dependencies are inferred from the claim text.